VulgarandStupid posted:

This guy did a thing.


https://youtu.be/IU0GDtj7FnM

forbidden dialectics posted:

My loop was almost $500 in fittings alone. Don't get into it for performance reasons; get into it if you're interested in it and treat it as a hobby.

CJacobs posted:

Canna Happy posted:

You need to know fan speeds, water/air temp deltas etc.

Cloner of the Elks posted:

That's a good point.

With that in mind, I just ordered:

• A bunch of Barrow fittings - 1x Ball valve, 2x 90 degree fittings, 2x 45 degree fittings, 6x standard fittings
  
• The Corsair Hydro X Series, XD5 RGB Pump/Reservoir Combo
  
• 1 litre Mayhems X1 premixed red coolant
  
• 1 litre Mayhems distilled water for pre-flushing the rad (this was also cheaper than any other distilled water I could find)
  
• 2x 500mm 10mm-ID, 14mm-OD PETG tubing. Wouldn't be surprised if I'll end up needing a bit more but this will do for now
  

Going to change my layout slightly from parallel to serial, so I think I have all the bits I need coming now. Will be interesting to see how this pans out.

And drat, I knew water cooling wasn't cheap but this is more than I expected.
Oh well, in for a penny, in for a pound!

Cloner of the Elks posted:

Ok, this will sound utterly stupid but hear me out!

Distilled water isn't as commonly available in UK as you might think. You can get it, but the price is utterly stupid when you finally do find it. Most of the 'closest' stuff you an find is Deionized water which from what I've read is not what should be used.

Distilled water on Amazon is, at the cheapest I can find with reasonable postage times, £7.90. That's 1.5 litres.

Mayhems distilled water with cleaning additives on Overclockers which is £2.50. that's 1 liter.

Quite why the price of evaporated water is so high, I have no idea. I'd make it myself if I had the time and desire to burn gas (though thinking about it, utilities would probably end up costing me more than £2.50 for less yield.

I spent more time looking for this goddamn water than I'll wilfully admit

Warmachine posted:

Anyway, EK just put out a Peltier. Just in case you want to bump your 300W Intel CPU up to 500W.

https://www.ekwb.com/shop/ek-quantumx-delta-tec

Cloner of the Elks posted:

Oh, and one more question though I suppose not specifically water cooling related.

Delidding.

Is it worth delidding and using liquid metal when water cooling? Will it make much of a difference at all or is it more for air cooled stuff?

I'm running a Kaby Lake i5 7600k and I'm toying with the idea but not sure if it's really worth it. I'd like to get the most out of my processor but it's not like it's struggling at standard clocked settings (I did plan to OC after the water cool build but have no specific numbers in mind)

Cloner of the Elks posted:

So I put my first fittings and tube in and stepped back to check out my work.



Then it hit me: Is the ball valve in a totally stupid place and essentially ineffective? I'm thinking that instead of coming off the side of a 3-way adapter it should go beneath, and a tube running to the rad below should be off the side instead?

PIZZA.BAT posted:

I'm looking at rebuilding my rig soon and am going to finally be doing away with my full tower and adopting SFF. I'm looking at Sliger cases and it looks like if I want to get *really* compact I'm limited to a 92mm AIO vs 240mm for some of their slightly larger cases. How much of a difference does this make? From my limited research it looks like the 92mm radiators are able to hold up just fine but I want to make sure I'm not missing anything

PIZZA.BAT posted:

Yeah after thinking about it a bit more I figured getting a slightly larger case which allows for 240mm radiators is worth the tradeoff.

DrDork posted:

If you're going Zen3 (and you should be) a 92mm AIO will be more than enough to keep it cool. I think people tend to over-estimate the heat output of things sometimes, and as Warmachine said, the difference between a 120mm and a 240mm is often like 5C, and frankly who cares about that unless it's running right up against a thermal throttle (and it won't be)? Like there is absolute no reason to care about your CPU having a load temp of 70C instead of 65C. It just doesn't matter.

The NR200 is a thread darling for a reason, though, so unless you really want the smallest case available, it's an excellent all around choice that won't have you hating yourself while you try to figure out how the hell you're going to route cables or whatever.

Nomyth posted:

Had another thought. Passively cool a CPU while dedicating a loop to the GPU for less noise while idle?

Indiana_Krom posted:

I have a custom water cooler, my EKWB D5 PWM pump is a little noisy at full throttle, but at 35% (~1875 RPM) it is totally silent and still circulates the entire volume of the loop in just a few seconds. Pump noise should never be a problem with a D5 as long as you get one with PWM control, same as fans, absolutely no reason to run them at 100% throttle pretty much ever. Loudest thing in my PC by miles is the 14 TB worth of mechanical hard drives for bulk storage, and they are only spinning about 5% of the time. My radiator is 420MM with 3x 140mm Noctua fans, case fans are 3x 140mm corsair fans, and both sets run at ~500 RPM and are also inaudible. The load temperature delta between 100% throttle on everything and normal silent settings is roughly 2C.

If you want silence, get a big radiator with big fans and run everything at fixed low RPMs. You can easily dump 400w worth of heat out of a radiator this size in total silence if you allow the coolant to run ~12-14C over ambient. It will yield roughly 50C GPU and 80C CPU temps when running respective power virus stress test loads (furmark and prime95 AVX at the same time).

Nomyth posted:

Ah, I misspoke. By passively cooling CPU I mean completely taking the CPU out of the loop altogether and using a (mostly) passive cooler on it, and dedicating all the radiator space in a case to GPU.

Nomyth posted:

Okay, it looks like then what I really need to do is address the source of my brainworms: what's a good way to prove to myself that I can turn down the fans and pump on what I have in my current system? The problem here is that I ended up putting together a ridiculously obscure Alphacool 280mm AIO monstrosity riding on my 9900K and my motherboard happens to be an equally obscure terrible choice (Supermicro) and I when I built this system spring of 2019 I hadn't really even given much thought at all to noise control because I stopped caring after playing around trying to get the all-core OC stable

really, the whole system was a poor choice and I need to find a place to listen to what a properly controlled loop looks/sounds like

Ak Gara posted:

Jesus christ lining the inside of my tower with black acrylic might have been a bad idea. What's a good anti static wipe that actually repels dust, but doesn't use alcohol? All I can find on amazon are essentially normal wet wipes.

DrDork posted:

Well I'm not really running a "recommended" loop here in the first place: I've got a 3080 + 5600X cooled by just a pair of 120mm rads. The pump is this Chinese special integrated CPU waterblock / pump / "res" deal, and runs at ~4000RPM constantly since apparently putting in a variable speed pump was within their budget, but allowing that pump speed to be controlled intelligently was not, so instead they just have a rheostat on a pair of wires to control the speed. And if that wasn't bad enough, I shoved it all in a Evolv shift. Yes, the type with the glass panels. Oh, and I don't have a water-temp sensor, but in retrospect that might not be the worst thing given the temp differentials.

So, yeah. Not a whole lot of rad space, hilariously poor case airflow, a "res" that's like...10ml? There's basically no spare fluid beyond what's in the tubes/blocks/rads. Frankly I'm kinda surprised this thing works at all. It's a monument to "you were so preoccupied with whether you could, you didn't stop to think if you should." (but it looks baller and it works so gently caress it)

Doing it this way means that the water gets pretty hot, which means that those rads actually do a lot of good work, because the higher the delta between your water and ambient, the more heat they can dump out. So I'd bet that the delta-T of the water before vs after the two rads is considerably higher than what it is in a more properly specced setup. The rads are non-directional, FWIW, as is the GPU block, so there was no opportunity to screw that up. The only thing that's directional at all is the pump, naturally.

But no joke, I've been running a buttcoin miner to help test the thermal abilities of this thing (because I can control how much GPU wattage it's using, anywhere between 200-320W, and because I can run it for long enough to heat-soak everything fully while still dicking around doing other stuff), and going GPU -> CPU had that poor Zen3 around 80-85 sustained, and 90C+ if it actually needed to clock up and do anything CPU-wise. Now that it's CPU -> GPU it's closer to 70C sustained, with no other changes--no need to remove the block from the CPU, since all I needed to do was swap the input and output feeds, which had enough slack in the lines already that nothing needed to be removed/replaced, just unplugged from one and moved over about 3cm.

I'm sure 75C water was an exaggeration. It's probably closer to 55C, since the GPU loads at 65C and the CPU at 75C. But yes, shoving 300W of heat output immediately before the CPU was very definitely causing higher CPU temps. If you run a more "sane" configuration where the water never heats up all that much in the first place, it probably doesn't make all that much of a difference. But with this thing, there absolutely are hotter/colder portions of the loop.

DrDork posted:

As noted, I don't have a coolant thermometer. The GPU temp I'm getting from Afterburner and the CPU temp from Ryzen Master. Those are, as far as things go, pretty authoritative sources for what those bits of hardware are running at. Even if the absolute numbers are somehow off, the fact that they changed after swapping the tubes to re-order things (while doing literally nothing else) remains valid.

Water temp I'm guessing at judging by how hot it still was ~10 min after turning it off so I could rearrange the tubes, at which point I'd ballpark it at 45C or so. Considering no pumps were running during that time, an operating load temp in the low 50C range seems perfectly reasonable.

DrDork posted:

Nope, I have a pair of 120mm rads. Specifically 120mm x 28mm, so they're on the thin side (otherwise only one fits). Behold my house of poor choices!

As far as "doing well," keeping a 320W GPU under 70C in a lovely case is already asking a fair amount from 240mm worth of rads. Throwing another 50-75W of CPU in there with a X570 chipset and some lights belching excess heat (why the hell are the UV lights that hot?) and I'm actually very pleasantly surprised.

I'll run the tests here in a few and get back to you two, but I'm not sure what you're looking for out of them: I'm pumping 300-400W of heat into a system with no effective res in a case well known for lovely airflow, and while 2x120 rads are keeping things below thermal limiting break points, they're obviously undersized for the task at hand (at least in the sense that everyone going into watercooling seems to obsess over keeping temps ultra low--this setup actually ends up being considerably quieter and cooler than the air setup beforehand, and they're no longer thermally limited, so in that sense the rads could be argued to be viably, if not optimally, sized for this).

ilkhan posted:

You should have warmer water and hotter component temps, but the water should still be consistent anywhere in the loop.

Input and output on which component(s) got switched? CPU blocks tend to be sensitive to in/out, but that's an extreme difference for just that change.

DrDork posted:

Input/Output for the CPU block / pump / res combo. Basically it's running backwards now. So before it was Res/Pump/CPU -> Rad -> Rad -> GPU -> Res/Pump/CPU, now it's Res/Pump/CPU <- Rad <- Rad <- GPU <- Res/Pump/CPU.

I think the high heat generation (~400W), low liquid volume, and low rad surface combines to exaggerate the impacts of ordering more than you'd see in a normal setup.

Truga posted:

i don't have rigid piping, but for my lovely loop it's normal that i have to refill the tank a little bit (it's a tiny tank) after a week, and then after a month of operation, as all the air bubbles filter out.

after that it ran for 4 years with no maintenance outside of removing and dusting the air filters twice a year

seriously considering going back to air right now tho. seeing the benches of newest cpus and gpus on water is super whatever. otoh i have a radeon so i could install a spicy bios...

Coredump posted:

I once had an almost silent build. But since then I’ve changed my mind from chasing silence to trying to change the nature of the sound to something I don’t mind. I like a regular fan in the background as some pleasant white noise. Just not coil whine or the sound of smaller fans.

Basically box fan with a car radiator is what I’m advocating for I guess.

Ak Gara posted:

I've tried to ignore that drat red wire, but I can't so went and brought some mesh cable wrap stuff.

Ak Gara posted:

dry location sue only
no-sfrviceable parts inside
for use withaudio video equifment
zone dangereuse
made in china

I'm sure it'll be fine!

Theophany posted:

Iirc Phobya were a fairly big brand back when watercooling was way more niche than it is today.


gently caress yeah.

ilkhan posted:

I know. But I try to block that craziness out of my mind as much as possible.

B-1.1.7 Bomber posted:

I have a question about ram blocks. I'm seeing a lot that are billed as designed for Corsair Dominator modules. They seem to be the only ram brand called out by name like this. How strict is this? I've got Crucial Ballistix. Do I want to avoid those blocks or will they also work with my stuff?

B-1.1.7 Bomber posted:

Right. Thanks. Most of these sites don't do a good job explaining these set-ups. Luckily there are quite a few to choose from, so I picked a "universal" style. And yes, I'm fully aware that ram waterblocks are totally unnecessary, but i'm all-in on this project.

B-1.1.7 Bomber posted:

I agree with you; it's weird that a couple of manufacturers call out Corsair Dominator sticks by name, and only this brand. Maybe they have a different profile that require different spreaders? Like many other things involving computers, I don't have a loving clue, but will forge ahead.